Temporal Context Mining for Learned Video Compression

This webinar will present the presenters work on end-to-end learned video compression, with a focus on temporal context mining. Their approach uniquely propagates both reconstructed frames and intermediate features to extract multi-scale temporal contexts, which are then effectively reused across the compression pipeline—including the contextual encoder-decoder, frame generator, and temporal context encoder. By replacing the computationally heavy auto-regressive entropy model, our codec achieves practical encoding/decoding speeds while delivering superior compression performance. Building on this foundation, this webinar will further discuss our advanced inter-prediction techniques specifically optimized for the low-delay scenario that achieve higher compression efficiency, and their novel extensions adapting this method for the random-access scenario.